brain-review-gate

Code review gates in a multi-agent workflow are enforced by validating worker output against audit requirements and beads state. This ensures unverified code integration is prevented by requiring strict adherence to signal-based decision making before task completion.

A beads-first invariant in multi-agent orchestration requires strict adherence to the beads system state to prevent invisible work. It validates that all worker output is backed by required audit sentinels and status transitions before integration into staging.

Audit trails for agent output are validated by checking worker diffs against required audit sentinels and status transitions. This process ensures the proposed changes are fully consistent with the project state before approving them for staging.

Yes, human-in-the-loop review is supported for multi-agent coding sessions where task completion requires strict evidence-based review. It detects scope drift and blocking issues to prevent unverified code from merging into staging branches.

Signal handling manages scope drift in multi-agent orchestration by detecting and managing critical risk flags before integration. This prevents incomplete agent work from being approved when scope drift or blocking issues are present in the worker output.

Automated review gates for agent output should not be used when the multi-agent workflow lacks beads state tracking or audit trail requirements. Without these prerequisites, signal-based decision making cannot properly validate worker output consistency.